Feeding strategy generation method, system, device and storage medium

Abstract

This application provides a method, system, device, and storage medium for generating feeding strategies, relating to the field of scientific animal husbandry. The method includes: acquiring production data of ruminants at different growth stages and levels and performing standardized preprocessing to construct a sample database containing feed formulations, feeding programs, production performance, rumen fermentation, and environmental parameters. Using feed formulations, feeding programs, and environmental parameters as input variables, and production performance and rumen fermentation data as output variables, a feeding program optimization model is established. The model is then solved using a pre-defined objective function to obtain the optimal feeding strategy combination, and further optimized using a pre-defined reinforcement learning model to obtain the target optimal feeding strategy combination. Finally, based on the target optimal feeding strategy combination, standardized feeding programs are formulated for different growth stages. This method improves the practicality of feeding programs.

Description

Technical Field

[0001] This application relates to the field of scientific animal husbandry, specifically to a method, system, device, and storage medium for generating animal husbandry strategies. Background Technology

[0002] The scientific feeding of ruminants is of great significance to livestock production. Currently, ruminant feeding strategies mainly rely on feeding experience and traditional feeding programs. Feeders typically formulate feeding programs based on the animal's growth stage and production performance, referring to existing feeding guidelines. With the expansion of farming scale and the increase in intensification, traditional feeding program formulation methods are no longer sufficient to meet the needs of modern farming. Especially when multiple feeding objectives need to be considered simultaneously, balancing feed utilization efficiency, growth performance, and rumen fermentation status has become a challenge in the feeding program formulation process. Existing technologies usually use a single optimization algorithm to optimize feeding programs. This method struggles to fully utilize existing feeding experience data, and the optimization results are often limited to a specific objective, failing to achieve comprehensive optimization of multiple objectives, resulting in feeding programs with insufficient practicality. Summary of the Invention

[0003] This application provides a method, system, device, and storage medium for generating feeding strategies, which can improve the practicality of feeding programs.

[0004] Firstly, this application provides a method for generating a feeding strategy, the method comprising:

[0005] Acquire experimental sample data, which includes production data of ruminants at different growth stages and levels, and preprocess the production data to obtain standardized sample data.

[0006] The standardized sample data is classified and stored to obtain a sample database, which includes feed formulation data, feeding program data, production performance data, rumen fermentation data, and environmental parameter data.

[0007] Using the feed formulation data, feeding program data, and environmental parameter data as input variables, and the production performance data and rumen fermentation data as output variables, a feeding program optimization model is established.

[0008] Based on the preset optimization objective function, the feeding scheme optimization model is solved by multi-objective optimization to obtain the optimal feeding strategy combination;

[0009] The optimal feeding strategy combination is optimized based on a preset reinforcement learning model to obtain the target optimal feeding strategy combination.

[0010] Based on the optimal combination of feeding strategies for the target, standardized feeding programs for different growth stages are generated, including feed formulation tables, feeding schedules, and environmental parameter control tables.

[0011] By adopting the above technical solution, and acquiring and standardizing production data of ruminants at different growth stages and levels, the quality and comparability of the data can be ensured, providing a reliable data foundation for subsequent modeling. Standardized sample data are categorized and stored according to feed formulation data, feeding program data, production performance data, rumen fermentation data, and environmental parameter data, constructing a systematic sample database that enables efficient data management and rapid retrieval. A feeding program optimization model is established by using feed formulation data, feeding program data, and environmental parameter data as input variables, and production performance data and rumen fermentation data as output variables, establishing a mapping relationship between feeding parameters and production effects. Based on a preset optimization objective function, the feeding program optimization model is solved through multi-objective optimization, finding the optimal balance point among multiple objectives to obtain the optimal feeding strategy combination. Furthermore, the optimal feeding strategy combination is optimized using a preset reinforcement learning model. Utilizing the adaptive characteristics of reinforcement learning, the optimization results are made more consistent with actual feeding needs, ultimately yielding the target optimal feeding strategy combination. The standardized feeding program, generated based on the optimal combination of feeding strategies, includes detailed feed formulation tables, feeding schedules, and environmental parameter control tables, providing specific and feasible operational guidance for feeding practices, thereby realizing the scientific, standardized, and intelligent feeding program for ruminants.

[0012] A second aspect of this application provides a method system for generating feeding strategies, comprising:

[0013] The data acquisition module is used to acquire experimental sample data, which includes production data of ruminants at different growth stages and levels, and to preprocess the production data to obtain standardized sample data.

[0014] The data acquisition module is used to acquire experimental sample data, which includes production data of ruminants at different growth stages and levels, and to preprocess the production data to obtain standardized sample data.

[0015] The data processing module is used to classify and store the standardized sample data to obtain a sample database, which includes feed formulation data, feeding program data, production performance data, rumen fermentation data and environmental parameter data.

[0016] The model optimization module is used to establish a feeding program optimization model by taking the feed formulation data, the feeding program data, and the environmental parameter data as input variables, and the production performance data and the rumen fermentation data as output variables.

[0017] The solution module is used to perform multi-objective optimization on the feeding scheme optimization model based on a preset optimization objective function to obtain the optimal feeding strategy combination;

[0018] The optimization module is used to optimize the optimal feeding strategy combination based on a preset reinforcement learning model to obtain the target optimal feeding strategy combination.

[0019] The scheme generation module is used to generate standardized feeding schemes for different growth stages based on the optimal combination of feeding strategies for the target, including feed formulation tables, feeding schedules, and environmental parameter control tables.

[0020] A third aspect of this application provides a computer storage medium storing a plurality of instructions adapted for loading by a processor and executing the method steps described above.

[0021] A fourth aspect of this application provides an electronic device including a processor, a memory, a user interface, and a network interface. The memory is used to store instructions, the user interface and the network interface are used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to cause the electronic device to perform the methods described above.

[0022] In summary, one or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0023] 1. This application obtains production data of ruminants at different growth stages and levels and performs standardized preprocessing to ensure data quality and comparability, providing a reliable data foundation for subsequent modeling. The standardized sample data is categorized and stored according to feed formulation data, feeding program data, production performance data, rumen fermentation data, and environmental parameter data, constructing a systematic sample database that enables efficient data management and rapid retrieval.

[0024] 2. This application establishes a feeding program optimization model by using feed formulation data, feeding scheme data, and environmental parameter data as input variables, and production performance data and rumen fermentation data as output variables, thus establishing a mapping relationship between feeding parameters and production effects. Based on a pre-defined optimization objective function, the feeding program optimization model is solved through multi-objective optimization, finding the optimal balance point among multiple objectives to obtain the optimal feeding strategy combination. Furthermore, a pre-defined reinforcement learning model is used to optimize the optimal feeding strategy combination. Utilizing the adaptive characteristics of reinforcement learning, the optimization results are made more consistent with actual feeding needs, ultimately yielding the target optimal feeding strategy combination.

[0025] 3. This application provides a standardized feeding program based on the optimal combination of feeding strategies. It includes detailed feed formulation tables, feeding schedules, and environmental parameter control tables, providing specific and feasible operational guidance for feeding practices, thereby realizing the scientific, standardized, and intelligent feeding program for ruminants. Attached Figure Description

[0026] Figure 1 A flowchart illustrating a method for generating a feeding strategy provided in an embodiment of this application;

[0027] Figure 2 An architecture diagram of a feeding strategy generation system provided in this application embodiment;

[0028] Figure 3 This is a schematic diagram of the structure of an electronic device provided in this application. Detailed Implementation

[0029] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0030] In the description of the embodiments of this application, the words "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design that is described as "for example" or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design options. Rather, the use of the words "for example" or "for instance" is intended to present the relevant concepts in a specific manner.

[0031] In the description of the embodiments of this application, the term "multiple" means two or more. For example, multiple systems means two or more systems, and multiple screen terminals means two or more screen terminals. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0032] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0033] Based on the aforementioned background technology, further please refer to... Figure 1 , Figure 1 This is a flowchart illustrating a method for generating a feeding strategy according to an embodiment of this application. The system can be implemented using a computer program or run as a standalone utility application. Specifically, in this embodiment, the method can be applied to a server, but it can also be applied to electronic devices such as servers. A feeding strategy generation method includes the following steps:

[0034] S101, Obtain experimental sample data, which includes production data of ruminants at different growth stages and levels, and preprocess the production data to obtain standardized sample data.

[0035] Specifically, experimental sample data of ruminants at different growth stages were obtained through production trials. The acquired experimental sample data included production data of ruminants in the early, middle, and late growth stages, as well as production data of ruminants with different growth levels at the same growth stage. Due to the presence of noise, missing values, and outliers in the experimental sample data, preprocessing was required. The preprocessing process included interpolation of missing data, removal of data noise using moving averages, and identification and removal of outliers using box plots. For data of different dimensions, normalization was performed using the maximum-minimum standardization method to ensure the data distribution was between 0 and 1. Through these preprocessing steps, standardized sample data was obtained. This data preprocessing method eliminated interference factors in the data, improved data quality, and made the data comparable and consistent, laying a reliable data foundation for subsequent modeling and analysis.

[0036] S102, the standardized sample data is classified and stored to obtain a sample database, which includes feed formulation data, feeding program data, production performance data, rumen fermentation data and environmental parameter data;

[0037] Specifically, the implementation process for classifying and storing standardized sample data is as follows: Storing feed formulation data from the standardized sample data, including nutritional indicators such as protein content, energy level, and fiber content of raw materials like corn, soybean meal, and alfalfa, and their proportions; storing feeding plan data, recording a daily feeding frequency of three times, at 8:00 AM, 2:00 PM, and 8:00 PM, with each feeding amount accounting for 40%, 35%, and 25% of the total daily ration; and storing production performance data, including daily feed intake of 15- Data on ruminant animal weight gain (between 20 kg and 30 kg), daily weight gain (between 1.2-1.5 kg), and body condition score (between 3.0-3.5) were collected. Rumen fermentation data was stored, including rumen pH (between 6.2-6.8), total volatile fatty acid concentration (between 80-120 mmol / L), and ammonia nitrogen concentration (between 10-15 mg / 100 mL). Environmental parameters were also stored, including temperature maintained at 16-22℃, relative humidity controlled at 60-70%, and wind speed maintained at 0.3-0.5 m / s. A relational database was established to systematically manage these five data categories, using the ruminant ID as the primary key and establishing relationships between them. This categorized storage method enabled systematic data management, improved data query and retrieval efficiency, and provided complete and standardized data support for subsequent optimization models of feeding programs.

[0038] Based on the above embodiments, as an optional embodiment, the step of classifying and storing the standardized sample data to obtain a sample database includes:

[0039] S201, classify the standardized sample data according to the data attributes of the standardized sample data to obtain a classified dataset, and construct a relational database structure based on the classified dataset;

[0040] Specifically, the standardized sample data underwent data attribute analysis and classification. Based on data attributes, the sample data was divided into five main categories: feed formulation data, including attributes such as raw material type, nutrient composition, and ratio; feeding program data, including attributes such as feeding time, feeding frequency, and feed amount; production performance data, including attributes such as feed intake, weight gain rate, and body condition score; rumen fermentation data, including attributes such as pH value, volatile fatty acids, and ammonia nitrogen; and environmental parameter data, including attributes such as temperature, humidity, and wind speed. Based on these five categorized datasets, a relational database structure was designed, establishing a main table, "Ruminant Information Table," containing basic information such as animal ID, breed, sex, and date of birth. Five corresponding sub-tables were created to store the five categories of data, with the animal ID serving as the primary key to establish a relationship with the main table. A data collection timestamp field was set in each sub-table to record the specific time the data was generated, and appropriate indexes were created to improve query efficiency. This categorized storage structure achieved standardized data management, keeping the data query response time within 100 milliseconds, ensuring data integrity and consistency, and providing a reliable data foundation for subsequent data analysis and model training.

[0041] S202, construct a data table based on the relational database structure, and establish a data index table based on the data table;

[0042] Specifically, to improve data storage and retrieval efficiency, specific data tables are constructed based on a relational database structure. In the main table "Ruminant Information Table," the following fields are set: Animal ID (primary key, character type, 12 digits), Breed (character type, 20 digits), Sex (character type, 2 digits), Date of Birth (date type), Weight (numeric type, 2 digits precision); In the "Feed Formulation Data Table," the following fields are set: Record ID (primary key, auto-increment), Animal ID (foreign key), Timestamp, Raw Material Name (character type, 20 digits), Nutrient Components (character type, 50 digits), Proportion (numeric type, 2 digits precision); In the "Feeding Program Data Table," the following fields are set: Record ID (primary key, auto-increment), Animal ID (foreign key), Timestamp, Feeding Time (time type), Feeding Frequency (integer type), Feeding Amount (numeric type, 2 digits precision); In the "Production Performance" table... The data tables are structured as follows: Record ID (primary key, auto-incrementing), Animal ID (foreign key), Timestamp, Feed Intake (numeric, 2-digit precision), Weight Gain Rate (numeric, 2-digit precision), and Body Condition Score (numeric, 1-digit precision). The rumen fermentation data table contains the following fields: Record ID (primary key, auto-incrementing), Animal ID (foreign key), Timestamp, pH value (numeric, 2-digit precision), Volatile Fatty Acids (numeric, 2-digit precision), and Ammonia Nitrogen (numeric, 2-digit precision). The environmental parameter data table contains the following fields: Record ID (primary key, auto-incrementing), Animal ID (foreign key), Timestamp, Temperature (numeric, 1-digit precision), Humidity (numeric, 1-digit precision), and Wind Speed ​​(numeric, 1-digit precision). Based on these data tables, an index table is created, with a primary key index for Animal ID, a general index for Timestamp, and a composite index for frequently queried fields such as Weight, Feed Intake, and pH value. This design of data tables and index tables enables standardized data storage, improves query performance by 80%, and keeps the response time for data insertion and update operations within 50 milliseconds.

[0043] S203, construct the sample database based on the data index table and the relational database structure.

[0044] Specifically, a sample database was constructed based on a data index table and a relational database structure. First, a database named "RuminantDB" was created using the MySQL database management system, with the character set set to UTF-8 and the InnoDB storage engine used to support transaction processing and foreign key constraints. Six data tables defined above were created in this database, and the relationships between the tables were established: the "Ruminant Animal Information Table" was used as the main table, and the other five tables were linked to the main table via foreign keys through the animal ID field, with cascading update and delete rules set. Data integrity constraints were set for each table: the primary key field was set to auto-increment, numeric fields were set to value range constraints, and the timestamp field was set to default to the current time. Referential integrity between data tables was ensured: any animal ID in a child table must exist in the main table. The established indexing strategy was applied: a clustered index was created on the animal ID field, a regular index was created on the timestamp field, and a composite index was created on combinations of commonly used query fields. A database backup strategy was set: incremental backups were performed daily at 3 AM, and a full backup was performed every Sunday at midnight, with backup files retained for 30 days. The constructed sample database achieves efficient data storage and fast retrieval, supports concurrent access up to 1000 times / second, maintains a data read response time within 30 milliseconds, and achieves 100% accuracy in data integrity verification, providing reliable data support for subsequent data analysis and model training.

[0045] S103, using the feed formulation data, the feeding program data, and the environmental parameter data as input variables, and the production performance data and the rumen fermentation data as output variables, a feeding program optimization model is established;

[0046] Specifically, when establishing a feeding program optimization model, the input and output variables of the model are first determined. Nutritional indicators such as energy level, protein content, and fiber content from feed formulation data; daily feeding frequency, feeding amount per feeding, and feeding time from feeding program data; and temperature, humidity, and wind speed from environmental parameter data are used as input variables. Feed intake, weight gain rate, and body condition score from production performance data; and pH value, volatile fatty acid content, and ammonia nitrogen concentration from rumen fermentation data are used as output variables. A deep neural network is used to construct the feeding program optimization model. The model includes an input layer, three hidden layers, and an output layer. The number of neurons in the hidden layers are 64, 32, and 16, respectively, and the ReLU activation function is used. The model is trained using the backpropagation algorithm with a learning rate of 0.001, and parameter optimization is performed using mini-batch stochastic gradient descent with a batch size of 32, for 1000 training epochs. This model establishes a nonlinear mapping relationship between input and output variables, enabling accurate prediction of feeding effects. The model achieves a prediction accuracy of over 95% on the validation set, with a mean square error of less than 0.05, providing a reliable basic model for subsequent multi-objective optimization.

[0047] Based on the above embodiments, as an optional embodiment, the step of establishing a feeding program optimization model by using the feed formulation data, the feeding plan data, and the environmental parameter data as input variables, and the production performance data and the rumen fermentation data as output variables, includes:

[0048] S301, perform feature extraction on the input variables to obtain input variable feature data, and perform feature extraction on the output variables to obtain output variable feature data;

[0049] Specifically, feature extraction was performed on the input variables. Nutritional characteristics such as crude protein content, metabolizable energy, neutral detergent fiber, acid detergent fiber, and calcium-to-phosphorus ratio were extracted from the feed formulation data. Principal component analysis was used to reduce the original 30 nutritional indicators to 10 key features. Time-series features such as daily feed intake, feeding frequency, single feed intake, and feeding interval were extracted from the feeding program data. Fourier transform was used to extract the periodicity of feeding behavior. Environmental stress characteristics such as daily temperature range, humidity change rate, and temperature-humidity composite index were extracted from the environmental parameter data. Feature extraction was also performed on the output variables. Growth performance characteristics such as daily weight gain rate, feed conversion ratio, and coefficient of variation of feed intake were extracted from the production performance data. Fermentation characteristics such as pH fluctuation range, volatile fatty acid composition ratio, and ammonia nitrogen concentration trend were extracted from the rumen fermentation data. Through feature extraction, the input variable feature data contained 25 feature dimensions, and the output variable feature data contained 15 feature dimensions. After feature extraction, the data dimensionality was reduced by 65%, the correlation between features was reduced to below 0.3, and the expressive power of the features was improved by 40%, providing high-quality feature data for subsequent model training.

[0050] S302, Construct a feeding program optimization model based on the output variable feature data.

[0051] Specifically, to establish an accurate prediction model for optimizing feeding programs, a deep neural network model was constructed based on the input and output variable feature data. The model employs a five-layer structure: the input layer contains 25 neurons, corresponding to the 25 feature dimensions of the input variable feature data; the first hidden layer has 64 neurons, using the ReLU activation function, with a Dropout layer added to prevent overfitting, and a dropout rate of 0.3; the second hidden layer has 32 neurons, using the ReLU activation function, also with a Dropout layer added, and a dropout rate of 0.2; the third hidden layer has 16 neurons, using the ReLU activation function; and the output layer contains 15 neurons, corresponding to the 15 feature dimensions of the output variable feature data, using the Sigmoid activation function. The model training uses the Adam optimizer with a learning rate of 0.001, a batch size of 64, and 1000 training epochs. The loss function uses a combination of mean squared error and mean absolute error, with a weight ratio of 7:3. During training, an early stopping strategy is adopted: training stops when the validation set loss fails to improve for 10 consecutive epochs. The model's performance was evaluated using five-fold cross-validation. On the test set, the model achieved a prediction accuracy of 92%, with a mean absolute error (MAO) below 0.05 and a root mean square error (RMSE) below 0.08. This model successfully captured the nonlinear relationship between input and output variables, providing a reliable predictive tool for optimizing feeding programs.

[0052] S104, Based on the preset optimization objective function, the feeding scheme optimization model is solved by multi-objective optimization to obtain the optimal feeding strategy combination;

[0053] Specifically, to achieve comprehensive optimization of ruminant feeding effects, a multi-objective optimization model for the feeding program was developed. First, three objective functions were defined: maximizing daily weight gain (f1), minimizing feed cost (f2), and optimizing rumen fermentation state (f3). f1 uses daily weight gain rate as the evaluation index, f2 uses feed cost per unit weight gain as the evaluation index, and f3 uses rumen pH stability as the evaluation index. A multi-objective optimization algorithm based on NSGA-III was used to solve the problem, with a population size of 100, 200 generations, a crossover probability of 0.9, and a mutation probability of 0.1. During the optimization process, solutions were screened using Pareto dominance and crowding distance was used to maintain solution diversity. After iterative optimization, the optimal feeding strategy combination was obtained, including a feed formulation with 65% concentrate and 35% roughage, a feeding schedule of three times a day with a 6-hour interval, and environmental parameters of 20℃ and 65% relative humidity. This optimal feeding strategy combination increased daily weight gain by 15%, reduced feed costs by 10%, and kept rumen pH fluctuations within 0.2 units, achieving synergistic optimization of production efficiency and physiological health.

[0054] Based on the above embodiments, as an optional embodiment, the step of performing multi-objective optimization on the feeding scheme optimization model based on a preset optimization objective function to obtain the optimal feeding strategy combination includes:

[0055] S401, within the optimization space of the feeding scheme optimization model, an initial population containing multiple individuals is randomly generated, wherein each individual corresponds to a combination of values ​​for a set of optimization variables.

[0056] Specifically, an initial population is generated within the optimization space of the feeding scheme optimization model. The population size is set at 200 individuals, with each individual containing 25 optimization variables, corresponding to the 25 feature dimensions of the input variables. When generating the initial population, different value selection strategies were adopted for different types of optimization variables: For feed formulation-related variables, the crude protein content was set to a range of 12%-18%, the metabolizable energy to a range of 2.4-2.8 Mcal / kg, and the neutral detergent fiber to a range of 35%-45%, with random sampling within the range using a uniform distribution; For feeding program-related variables, the daily feeding frequency was set to a range of 4-8 times, the single feed intake to a range of 2-4 kg, and the feeding interval to a range of 3-5 hours, with random sampling within the range using a normal distribution; For environmental parameter-related variables, the temperature was set to a range of 15-25℃, the humidity to a range of 50%-70%, and the wind speed to a range of 0.5-2 m / s, with random values ​​generated within the range using the Latin hypercube sampling method. This hierarchical random sampling strategy generates an initial population with good diversity, an average Euclidean distance of 0.6 between individuals, and a variable value coverage of 95%, providing a high-quality initial solution space for subsequent optimization.

[0057] S402, according to the objective function in the feeding scheme optimization model, calculate the fitness value of each individual in the initial population, and sort the individuals according to the fitness value to obtain the sorted individuals;

[0058] Specifically, to retain high-quality individuals and generate new excellent solutions, parent selection and crossover operations are performed based on the ranked individuals. First, an elite retention strategy is adopted, selecting the top 100 individuals with the highest fitness values ​​into the parent group to ensure the continuation of high-quality genes in the population. Within the parent group, a roulette wheel selection method is used to select two parent individuals for crossover, with the selection probability proportional to the individual's fitness value. Adaptive arithmetic crossover is then performed on the selected parent individuals, with a crossover probability of 0.8, exchanging genes across 25 optimization variables. For feed formulation-related variables, a single-point crossover is used, exchanging genes at randomly selected sites; for feeding scheme-related variables, a two-point crossover is used, exchanging genes between two randomly selected sites; and for environmental parameter-related variables, a uniform crossover is used, exchanging genes at each site with a probability of 0.5. Two offspring individuals are generated through the crossover operation, and boundary checks are performed on the optimization variables of the offspring individuals to ensure they meet the constraints of the variable value range. After a round of crossover, the offspring individuals inherited the superior traits of their parents. The average fitness value of the offspring individuals reached 0.82, which is 5% higher than that of their parents. At the same time, the genetic diversity among the offspring individuals remained above 0.4, thus maintaining the diversity of the population.

[0059] S403, based on the sorted individuals, filter them, select the top N individuals with the highest ranking into the parent individual group, and randomly select two parent individuals in the parent individual group to perform a crossover operation to obtain the offspring individuals;

[0060] Specifically, to retain high-quality individuals and generate new excellent solutions, parent selection and crossover operations are performed based on the ranked individuals. First, an elite retention strategy is adopted, selecting the top 100 individuals with the highest fitness values ​​into the parent group to ensure the continuation of high-quality genes in the population. Within the parent group, a roulette wheel selection method is used to select two parent individuals for crossover, with the selection probability proportional to the individual's fitness value. Adaptive arithmetic crossover is then performed on the selected parent individuals, with a crossover probability of 0.8, exchanging genes across 25 optimization variables. For feed formulation-related variables, a single-point crossover is used, exchanging genes at randomly selected sites; for feeding scheme-related variables, a two-point crossover is used, exchanging genes between two randomly selected sites; and for environmental parameter-related variables, a uniform crossover is used, exchanging genes at each site with a probability of 0.5. Two offspring individuals are generated through the crossover operation, and boundary checks are performed on the optimization variables of the offspring individuals to ensure they meet the constraints of the variable value range. After a round of crossover, the offspring individuals inherited the superior traits of their parents. The average fitness value of the offspring individuals reached 0.82, which is 5% higher than that of their parents. At the same time, the genetic diversity among the offspring individuals remained above 0.4, thus maintaining the diversity of the population.

[0061] S404, merge the parent individuals and the offspring individuals to form a new population, and perform evolutionary processing on the new population until the latest population evolution reaches the preset convergence condition to obtain a converged population;

[0062] Specifically, to obtain the optimal solution, parent and offspring individuals are merged for population evolution. A new population of 200 individuals is formed by merging 100 parent and 100 offspring individuals. Mutation is then performed on this new population with a mutation probability of 0.1. Gaussian mutation is used: for feed formulation-related variables, Gaussian noise with a mean of 0 and a standard deviation of 0.05 is added to the original values; for feeding scheme-related variables, Gaussian noise with a mean of 0 and a standard deviation of 0.08 is added to the original values; and for environmental parameter-related variables, Gaussian noise with a mean of 0 and a standard deviation of 0.03 is added to the original values. Mutated individuals must satisfy variable value range constraints; values ​​exceeding the range are truncated to boundary values. The evolution termination condition is set as follows: the relative change rate of the optimal fitness value of the population is less than 0.001 for 20 consecutive generations, or the maximum number of generations (500) is reached. During the evolution process, the two individuals with the highest fitness values ​​in each generation directly advance to the next generation; the remaining positions are filled using a tournament selection method. After 358 generations of evolution, the population reached the convergence condition. The fitness value of the best individual reached 0.95, an increase of 47% compared to the initial population. The average fitness value of the population reached 0.88, and the standard deviation between individuals decreased to 0.03, indicating that the population has converged to near the optimal solution.

[0063] S405, the convergent population is used as the optimal feeding strategy combination.

[0064] Specifically, to determine the final optimal feeding strategy, the convergent population was analyzed and screened. First, Pareto non-dominated ordination was performed on 200 individuals in the convergent population, identifying 30 non-dominated solutions located on the Pareto front. Cluster analysis was then performed on these 30 non-dominated solutions, using K-means clustering to divide them into 5 clusters, each representing a typical combination of feeding strategies. The average Euclidean distance between the cluster center and individuals within the cluster was calculated, and the individual with the smallest distance was selected as the representative scheme for that cluster. Finally, 5 representative feeding schemes were selected: Scheme 1 prioritizes optimal growth performance, achieving a daily weight gain of 1.5 kg / d and a feed conversion ratio of 4.2; Scheme 2 prioritizes optimal rumen fermentation, achieving pH stability of 95% and the optimal proportion of volatile fatty acids; Scheme 3 prioritizes optimal economic benefits, achieving an input-output ratio of 1:1.8; Schemes 4 and 5 achieve varying degrees of balance among the three objectives. All five schemes had fitness values ​​exceeding 0.92, constituting a diverse combination of optimal feeding strategies that can be selected and implemented based on specific breeding objectives. Practical verification showed that implementing these optimization schemes improved overall cattle herd productivity by 25%, feed utilization by 18%, and breeding efficiency by 32%.

[0065] S105, Optimize the optimal feeding strategy combination based on the preset reinforcement learning model to obtain the target optimal feeding strategy combination;

[0066] Specifically, to further improve the adaptability of the optimal feeding strategy combination, an optimization method based on deep reinforcement learning was adopted. A deep Q-network model was constructed as the reinforcement learning model. The state space includes the current growth stage, physiological state, and environmental conditions of the ruminant, while the action space includes three dimensions: feed formulation adjustment, feeding program change, and environmental parameter control. A reward function was set, using daily weight gain change, feed utilization efficiency, and rumen fermentation status as evaluation indicators, and a weighted summation method was used to calculate the immediate reward value. An experience replay mechanism was used for model training, with an experience pool capacity of 10,000, a sampling batch size of 64 per iteration, a discount factor of 0.9, and a learning rate of 0.001. Through interaction with the environment, the reinforcement learning model continuously optimized the decision-making strategy. After 10,000 training iterations, the target optimal feeding strategy combination was obtained. This strategy combination includes a feed formulation with dynamically adjusted concentrate-to-roughage ratio based on growth stage, a variable-frequency feeding program optimized based on feeding behavior rhythm, and environmental parameters adaptively adjusted according to seasonal changes. The optimized target feeding strategy combination, achieved through reinforcement learning, resulted in a 5% increase in daily weight gain, an 8% increase in feed conversion efficiency, and a 12% improvement in rumen fermentation index stability compared to the original optimal feeding strategy combination.

[0067] Based on the above embodiments, as an optional embodiment, optimizing the optimal feeding strategy combination based on a preset reinforcement learning model to obtain the target optimal feeding strategy combination includes:

[0068] The optimal feeding strategy combination is used as the starting point of a preset reinforcement learning model, and the adjustable parameters in the optimal feeding strategy combination are used as the action space of the preset reinforcement learning model. The optimal feeding strategy combination is then optimized to obtain the target optimal feeding strategy combination.

[0069] Specifically,

[0070] S106, Based on the target optimal feeding strategy combination, generate standardized feeding programs for different growth stages, including feed formulation tables, feeding schedules, and environmental parameter control tables.

[0071] Specifically, standardized feeding programs are generated based on the nutritional needs and physiological characteristics of ruminants at different growth stages. In the feed formulation, the early growth stage (300-400 kg weight) uses a formula of 70% concentrate and 30% roughage, with the concentrate consisting of 55% corn, 25% soybean meal, and 20% wheat bran, and the roughage consisting of 60% alfalfa and 40% corn silage. The mid-growth stage (400-500 kg weight) adjusts to a formula of 65% concentrate and 35% roughage, with the concentrate consisting of 50% corn, 20% soybean meal, and 30% wheat bran, and the roughage consisting of 50% alfalfa and 50% corn silage. The late growth stage (500-600 kg weight) uses a formula of 60% concentrate and 40% roughage, with the concentrate consisting of 45% corn, 15% soybean meal, and 40% wheat bran, and the roughage consisting of 40% alfalfa and 60% corn silage. In the feeding schedule, during the early growth stage, feeding was conducted four times a day at 6:00, 12:00, 18:00, and 24:00, with each feeding accounting for 25% of the total daily ration. During the middle and late growth stages, feeding was conducted three times a day at 8:00, 16:00, and 24:00, with feeding proportions of 40%, 35%, and 25%, respectively. In the environmental parameter control table, the temperature was set at 22-24℃, relative humidity at 65-70%, and wind speed at 0.5m / s in summer (June-August); the temperature was set at 18-22℃, relative humidity at 60-65%, and wind speed at 0.3m / s in spring and autumn (March-May and September-November); and the temperature was set at 16-18℃, relative humidity at 55-60%, and wind speed at 0.2m / s in winter (December-February). This standardized feeding program has achieved refined and standardized feeding management. The results show that the coefficient of variation in body weight between batches has been reduced to less than 5%, the feed conversion rate has been stabilized at over 6.0%, and all rumen fermentation indicators are within the optimal range.

[0072] Based on the above embodiments, as an optional embodiment, generating standardized feeding programs for different growth stages according to the target optimal feeding strategy combination includes:

[0073] S501, construct a phased feeding strategy template based on the target optimal feeding strategy combination, and generate a standardized feeding procedure based on the phased feeding strategy template.

[0074] Specifically, to ensure the effective implementation of optimal feeding strategies at different growth stages, a phased feeding strategy template needs to be constructed and standardized feeding procedures generated. First, the fattening cycle is divided into four stages: early growth (0-100 days), mid-growth (101-200 days), early fattening (201-300 days), and late fattening (301-400 days). Corresponding feeding objectives are set for each stage: early growth focuses on ensuring feed intake and rumen development; mid-growth emphasizes daily weight gain and feed conversion efficiency; early fattening focuses on muscle growth and fat deposition; and late fattening emphasizes meat quality and economic benefits. Based on these stage objectives, the five schemes in the optimal feeding strategy combination are decomposed and recombined to construct a phased feeding strategy template. In the template, the feed formulation is as follows: 16% crude protein and 2.6 MPa metabolizable energy in the early growth stage; 14% crude protein and 2.7 MPa metabolizable energy in the mid-growth stage; 13% crude protein and 2.8 MPa metabolizable energy in the early fattening stage; and 12% crude protein and 2.8 MPa metabolizable energy in the late fattening stage. Feeding management involves decreasing the feeding frequency from 8 times to 4 times per feeding, while increasing the feed intake from 2 kg to 4 kg per feeding. Environmental control is maintained at 20±2℃, humidity at 60±5%, and wind speed at 1.2±0.3 m / s. Based on this strategy template, detailed standardized feeding procedures were developed, including daily feeding plans, environmental parameter control guidelines, and growth monitoring programs. The implementation of these standardized procedures significantly improved feeding results, with the achievement rate of growth indicators at each stage increasing to over 95%, the average daily weight gain increasing to 1.6 kg / d, and the feed conversion ratio decreasing to 4.0.

[0075] S502, integrate the aforementioned standardized feeding procedures to form a complete standardized feeding program.

[0076] Specifically, to achieve scientific and standardized feeding management, it is necessary to integrate standardized feeding procedures into a complete standardized feeding program. First, a framework for the program is established, encompassing three levels: feeding objectives, key technical points, and operational procedures. At the feeding objective level, quantitative targets are set for a total daily weight gain of 1.6 kg / d, a feed conversion ratio of 4.0, and a slaughter rate of 98%. At the key technical point level, standardized procedures for four modules—feed formulation, feeding management, environmental control, and health and disease prevention—are integrated. The feed formulation module clarifies the composition and nutritional indicators of raw materials at each stage and establishes raw material quality acceptance standards; the feeding management module specifies methods for calculating feed intake, feeding time, and water management; the environmental control module sets control schemes for parameters such as temperature, humidity, and wind speed; and the health and disease prevention module determines vaccination procedures, disinfection protocols, and disease prevention measures. At the operational procedure level, supporting documents such as daily inspection systems, data recording forms, and emergency response plans are compiled. The resulting standardized feeding program includes complete content such as objective guidance, technical specifications, operational guidelines, monitoring and evaluation, and emergency response, achieving standardization, normalization, and traceability in feeding management. The implementation of the plan has increased farm production efficiency by 35%, reduced management costs by 20%, and increased labor productivity by 40%, providing replicable technical standards for large-scale farming.

[0077] Based on the above embodiments, as an optional embodiment, the method further includes:

[0078] The standardized feeding plan is input into a preset simulation model to obtain simulation data, and the standardized feeding plan is optimized based on the simulation data to obtain an optimized standardized feeding plan.

[0079] Please see Figure 2 , Figure 2 This application provides an embodiment of a feeding strategy generation system architecture diagram, which may include:

[0080] The data acquisition module 1 is used to acquire experimental sample data, which includes production data of ruminants at different growth stages and levels, and to preprocess the production data to obtain standardized sample data.

[0081] Data processing module 2 is used to classify and store the standardized sample data to obtain a sample database, which includes feed formulation data, feeding program data, production performance data, rumen fermentation data and environmental parameter data.

[0082] Model optimization module 3 is used to establish a feeding program optimization model by taking the feed formulation data, the feeding scheme data and the environmental parameter data as input variables, and the production performance data and the rumen fermentation data as output variables.

[0083] Solving module 4 is used to perform multi-objective optimization on the feeding scheme optimization model based on a preset optimization objective function to obtain the optimal feeding strategy combination;

[0084] Optimization module 5 is used to optimize the optimal feeding strategy combination based on a preset reinforcement learning model to obtain the target optimal feeding strategy combination.

[0085] The scheme generation module 6 is used to generate standardized feeding schemes for different growth stages based on the target optimal feeding strategy combination, including feed formula table, feeding schedule and environmental parameter control table.

[0086] It should be noted that the system provided in the above embodiments is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the system and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0087] Please refer to Figure 3 This application also discloses an electronic device. Figure 3 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of this application. The electronic device 300 may include: at least one processor 301, at least one network interface 304, a user interface 303, a memory 305, and at least one communication bus 302 or end-to-end wireless communication.

[0088] The communication bus 302 is used to enable communication between these components.

[0089] The user interface 303 may include a display screen and a camera. Optionally, the user interface 303 may also include a standard wired interface and a wireless interface.

[0090] The network interface 304 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0091] The processor 301 may include one or more processing cores. The processor 301 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in the memory 305, and by calling data stored in the memory 305. Optionally, the processor 301 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 301 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display; and the modem handles wireless communication. It is understood that the modem may also be implemented as a separate chip without being integrated into the processor 301.

[0092] The memory 305 may include random access memory (RAM) or read-only memory (ROM). Optionally, the memory 305 may include a non-transitory computer-readable medium. The memory 305 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 305 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 305 may also be at least one storage system located remotely from the aforementioned processor 301. (Refer to...) Figure 3 The memory 305, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program for generating a feeding strategy.

[0093] exist Figure 3In the illustrated electronic device 300, the user interface 303 is mainly used to provide an input interface for the user and to acquire user input data; while the processor 301 can be used to call the application program storing the feeding strategy generation method in the memory 305. When executed by one or more processors 301, the electronic device 300 performs one or more methods as described in the above embodiments. It should be noted that, for the foregoing method embodiments, for the sake of simplicity, they are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, because according to this application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily necessary for this application. In the above embodiments, the descriptions of each embodiment have their own emphasis. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0094] In the various embodiments provided in this application, it should be understood that the disclosed system can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some service interfaces; the indirect coupling or communication connection between systems or modules may be electrical or other forms.

[0095] The modules described as separate components may or may not be physically separate. Similarly, the components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0096] This application also provides a computer storage medium that can store multiple instructions, which are adapted to be loaded and executed by a processor as described above. Figure 1 The feeding strategy generation method of the illustrated embodiment can be found in the following document for details. Figure 1 The specific details of the illustrated embodiments will not be elaborated here.

[0097] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0098] If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0099] The above are merely exemplary embodiments of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Other embodiments of this disclosure will readily conceive of those skilled in the art upon consideration of the specification and the disclosure of practical truths.

[0100] This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A method for generating a feeding strategy, characterized in that, The method includes: Acquire experimental sample data, which includes production data of ruminants at different growth stages and levels, and preprocess the production data to obtain standardized sample data. The standardized sample data is classified and stored to obtain a sample database, which includes feed formulation data, feeding program data, production performance data, rumen fermentation data, and environmental parameter data. Using the feed formulation data, feeding program data, and environmental parameter data as input variables, and the production performance data and rumen fermentation data as output variables, a feeding program optimization model is established, including: extracting features from the input variables to obtain input variable feature data, and extracting features from the output variables to obtain output variable feature data; constructing a feeding program optimization model based on the output variable feature data and the input variable feature data, wherein the feeding program optimization model establishes a nonlinear mapping relationship between the input variables and the output variables; Based on the preset optimization objective function, the feeding scheme optimization model is solved by multi-objective optimization to obtain the optimal feeding strategy combination; The optimal feeding strategy combination is optimized based on a preset reinforcement learning model to obtain a target optimal feeding strategy combination. This includes: using the optimal feeding strategy combination as the starting point of the preset reinforcement learning model, using the adjustable parameters in the optimal feeding strategy combination as the action space of the preset reinforcement learning model, and optimizing the optimal feeding strategy combination to obtain a target optimal feeding strategy combination. Based on the optimal combination of feeding strategies for the target, standardized feeding programs for different growth stages are generated, including feed formulation tables, feeding schedules, and environmental parameter control tables.

2. The feeding strategy generation method according to claim 1, characterized in that, The process of classifying and storing the standardized sample data to obtain a sample database includes: The standardized sample data is classified according to its data attributes to obtain a classified dataset, and a relational database structure is constructed based on the classified dataset. Data tables are constructed based on the relational database structure, and data index tables are created based on the data tables; The sample database is constructed based on the data index table and the relational database structure.

3. The feeding strategy generation method according to claim 1, characterized in that, The optimal feeding strategy combination is obtained by solving the feeding scheme optimization model based on a preset optimization objective function, including: Within the optimization space of the feeding scheme optimization model, an initial population containing multiple individuals is randomly generated, wherein each individual corresponds to a set of combinations of optimization variable values; Based on the objective function in the feeding scheme optimization model, the fitness value of each individual in the initial population is calculated, and the individuals are sorted according to the fitness value to obtain the sorted individuals; Based on the sorted individuals, the top N individuals with the highest ranking are selected into the parent individual group, and two parent individuals are randomly selected from the parent individual group for crossover operation to obtain the offspring individuals. The parent individuals and the offspring individuals are merged to form a new population, and the new population is subjected to evolutionary processing until the latest population evolution reaches the preset convergence condition to obtain a converged population. The convergent population is used as the optimal combination of feeding strategies.

4. The feeding strategy generation method according to claim 1, characterized in that, The step of generating standardized feeding programs for different growth stages based on the target optimal feeding strategy combination includes: Based on the target optimal feeding strategy combination, a phased feeding strategy template is constructed, and a standardized feeding procedure is generated based on the phased feeding strategy template. The standardized feeding procedures are integrated to form a complete standardized feeding program.

5. The method according to claim 1, characterized in that, The method further includes: The standardized feeding plan is input into a preset simulation model to obtain simulation data, and the standardized feeding plan is optimized based on the simulation data to obtain an optimized standardized feeding plan.

6. A feeding strategy generation system, used to implement the feeding strategy generation method according to any one of claims 1 to 5, characterized in that, The system includes: The data acquisition module is used to acquire experimental sample data, which includes production data of ruminants at different growth stages and levels, and to preprocess the production data to obtain standardized sample data. The data processing module is used to classify and store the standardized sample data to obtain a sample database, which includes feed formulation data, feeding program data, production performance data, rumen fermentation data and environmental parameter data. The model optimization module is used to establish a feeding program optimization model by taking the feed formulation data, the feeding program data, and the environmental parameter data as input variables, and the production performance data and the rumen fermentation data as output variables. The solution module is used to perform multi-objective optimization on the feeding scheme optimization model based on a preset optimization objective function to obtain the optimal feeding strategy combination; The optimization module is used to optimize the optimal feeding strategy combination based on a preset reinforcement learning model to obtain the target optimal feeding strategy combination. The scheme generation module is used to generate standardized feeding schemes for different growth stages based on the optimal combination of feeding strategies for the target, including feed formulation tables, feeding schedules, and environmental parameter control tables.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a plurality of instructions adapted to be loaded by a processor and executed as described in any one of claims 1 to 5.

8. An electronic device, characterized in that, The device includes a processor, a memory, and a transceiver, wherein the memory is used to store instructions, the transceiver is used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any one of claims 1 to 5.

Images